large Correlation sets and predictors.

Consider the Cross Correlations for Australian end of day data for the combined:

• ASX Equities
• Australian indices
• Commodities
• Continuous futures
• Economic indicators,
• AU forex
• World indices

These together form a set of 7000 EOD series.

A simple cross correlation would involve 7000*7000/2 cross correlations =2.45*10^7 correlations

Then add in a variable window size for the correlation going form 5 days to 2 years. Roughly increasing correlations to be performed by 500, leaving 2.45 * 10^7 = 1.2*10^10 correlations.

Then add a variable lag from 0 days to 180 days, again increasing the number of correlations by 180

Leading to 180* 1.2*10^10 correlations or approximately 2*10^12 correlations
or roughly a 2 trillion correlations.

Then check to see if there is any statistical significance with null hypothesis testings.

I was wondering if there are any of the shelf finance/trading platforms that can do this or if anyone has done that.

And what did they find in terms of significance.

Also what platform/tool did they use.

And if they did this did the use this in a predictor matrix.